Lightweight UAV Borne 3D Perception via Pan-Tilt 2D LiDAR

Light detection and ranging (LiDAR) provides accurate and lighting-invariant 3D maps essential for autonomous Unmanned Aerial Vehicle (UAV) navigation, localization, and obstacle avoidance. However, conventional 3D LiDARs exceed the size, weight, power, and cost constraints (SWaP-C) of small UAVs. This work presents a low-SWaP 3D perception system that converts an inexpensive 2D LiDAR into a volumetric mapper and obstacle detector using servo-driven pan-tilt actuation. The proposed system generates dense 3D point clouds in real time and detects obstacles using a lightweight perception pipeline. The system employs a containerized ROS architecture designed for deployment on embedded platforms with limited computational resources. Bench-top and flight experiments on a custom UAV demonstrate full panoramic mapping in 8-12 seconds with average processing latency under 114 ms, validating the system's suitability for airborne 3D sensing applications.